From ca65fa9a7f3d72d62ce980ee16b3b20ed7bdedef Mon Sep 17 00:00:00 2001
From: Nikita Nemkin <nikita@nemkin.ru>
Date: Fri, 14 Feb 2025 22:13:18 +0500
Subject: [PATCH] string: Allow references to unmatched groups in REGEX REPLACE

References to unmatched groups will be replaced with empty strings.

Issue: #26629
Fixes: #19012
---
 Help/command/string.rst                 | 3 +++
 Help/release/dev/regex-fixes.rst        | 3 +++
 Source/cmStringReplaceHelper.cxx        | 6 ++----
 Tests/CMakeTests/StringTest.cmake.in    | 2 +-
 Tests/CMakeTests/StringTestScript.cmake | 4 ++++
 5 files changed, 13 insertions(+), 5 deletions(-)

diff --git a/Help/command/string.rst b/Help/command/string.rst
index c510ff44d5..b125e4b975 100644
--- a/Help/command/string.rst
+++ b/Help/command/string.rst
@@ -122,6 +122,9 @@ Search and Replace With Regular Expressions
   string instead of the beginning of each repeated search.
   See policy :policy:`CMP0186`.
 
+  The replacement expression may contain references to subexpressions that
+  didn't match anything. Previously, such references triggered an error.
+
 .. _`Regex Specification`:
 
 Regex Specification
diff --git a/Help/release/dev/regex-fixes.rst b/Help/release/dev/regex-fixes.rst
index e979c03efe..67f80ed6c4 100644
--- a/Help/release/dev/regex-fixes.rst
+++ b/Help/release/dev/regex-fixes.rst
@@ -3,3 +3,6 @@ regex-fixes
 
 * Regular expressions match the ``^`` anchor at most once in repeated
   searches, at the start of the input.  See policy :policy:`CMP0186`.
+
+* References to unmatched groups are allowed, they are replaced with empty
+  strings.
diff --git a/Source/cmStringReplaceHelper.cxx b/Source/cmStringReplaceHelper.cxx
index 5cd159e235..e909b05a31 100644
--- a/Source/cmStringReplaceHelper.cxx
+++ b/Source/cmStringReplaceHelper.cxx
@@ -61,10 +61,7 @@ bool cmStringReplaceHelper::Replace(std::string const& input,
       } else {
         // Replace with part of the match.
         auto n = replacement.Number;
-        auto start = this->RegularExpression.start(n);
-        if (start != std::string::npos) {
-          output += this->RegularExpression.match(n);
-        } else {
+        if (n > this->RegularExpression.num_groups()) {
           std::ostringstream error;
           error << "replace expression \"" << this->ReplaceExpression
                 << "\" contains an out-of-range escape for regex \""
@@ -72,6 +69,7 @@ bool cmStringReplaceHelper::Replace(std::string const& input,
           this->ErrorString = error.str();
           return false;
         }
+        output += this->RegularExpression.match(n);
       }
     }
 
diff --git a/Tests/CMakeTests/StringTest.cmake.in b/Tests/CMakeTests/StringTest.cmake.in
index 6a94cc5852..874655130f 100644
--- a/Tests/CMakeTests/StringTest.cmake.in
+++ b/Tests/CMakeTests/StringTest.cmake.in
@@ -84,7 +84,7 @@ check_cmake_test(String
 # Execute each test listed in StringTestScript.cmake:
 #
 set(scriptname "@CMAKE_CURRENT_SOURCE_DIR@/StringTestScript.cmake")
-set(number_of_tests_expected 72)
+set(number_of_tests_expected 73)
 
 include("@CMAKE_CURRENT_SOURCE_DIR@/ExecuteScriptTests.cmake")
 execute_all_script_tests(${scriptname} number_of_tests_executed)
diff --git a/Tests/CMakeTests/StringTestScript.cmake b/Tests/CMakeTests/StringTestScript.cmake
index 7c4585791b..dc7b8ae61f 100644
--- a/Tests/CMakeTests/StringTestScript.cmake
+++ b/Tests/CMakeTests/StringTestScript.cmake
@@ -116,6 +116,10 @@ elseif(testname STREQUAL regex_replace_index_too_small) # fail
 elseif(testname STREQUAL regex_replace_index_too_large) # fail
   string(REGEX REPLACE "^this (.*)$" "with \\1 \\2" v "this input")
 
+elseif(testname STREQUAL regex_replace_index_no_match) # pass
+  string(REGEX REPLACE "^(this (.*)|(that .*))$" "with \\1 \\2 \\3" v "this input")
+  message(STATUS "v='${v}'")
+
 elseif(testname STREQUAL compare_no_mode) # fail
   string(COMPARE)